# RE: st: How to test the equality of coef in regression model

 From "David Harrison" <[email protected]> To <[email protected]> Subject RE: st: How to test the equality of coef in regression model Date Mon, 24 Apr 2006 11:37:35 +0100

```The -cnsreg- approach is numerically identical to generating the new
variable x as suggested in ronggui's original post, but looks rather
more complicated to me.

The only error in ronggui's original code seems to be in the
specification of the two -reg- commands, which should be:

reg y x1 x2 x3
reg y x

These two models are nested.

David

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Joseph
Coveney
Sent: 24 April 2006 09:45
To: Statalist
Subject: Re: st: How to test the equality of coef in regression model

ronggui wrote:

First,I get the first model by:
reg y x1 x2 x2 x3

then I want to test if _b[x1]=_b[x2]=_b[x3], Can I achieve it by the
following codes?
gen x=x1+x2+x3
reg y x1 x2 x2 x3
est store A1
reg y x1 x
lrtest A1 .

Are model A1 and . nested? and Is this test valid?

Thanks.

------------------------------------------------------------------------
--------

Maybe you could use constrained regression (-help constraint- and -help
cnsreg-) or just do a Wald test (-help test-).  They both seem to arrive
at
the same conclusion in the example below.

Joseph Coveney

clear
set more off
set obs 25
set seed `=date("2006-04-24", "ymd")'
forvalues i = 1/3 {
generate float x`i' = uniform()
}
generate float y = 2 * x1 + 2 * x2 + 2 * x3 + invnorm(uniform())
regress y x*
test x1 = x2, notest
test x2 = x3, accumulate
estimates store A
constraint 1 x1 = x2
constraint 2 x2 = x3
cnsreg y x*, constraints(1-2)
lrtest A ., force
exit

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
______________________________________________________________________

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.